In this specific case the user is looking at a two-column strategy – (year, period) – which has a big impact on the choice of best partitioning strategy.

Remember that I asked the question: ‘Do you not have to deal with queries like “data between year 2010 period 12 and year 2011 period 2″ – if so how will they work ?’ and got the answer: ‘This should not happen. And even if they do it will just end up looking at two partitions.. If we take the OP at face value then there’s a good reason for looking at all possible solutions.

List partitioning is good on the year because there are only a few distinct values, not a continuous range of values. The same argument applies to list partitioning on period.

List partitions do (to a limited extent) allow for range-based predicates. I haven’t done any experiments on list/list composites, but on a simple list partition you can have a predicate like: “year between 2008 and 2010″ and the optimizer will recognise this as an opportunity for an inlist iterator and take advantage of partition pruning.